Internet Penetration: A New Explainable and Highly Accurate Predictor of Worldwide Cancer Incidence by Dr Chris Barnes, Bangor Scientific and Educational Consultants, Gwynedd, Wales. E-mail email@example.com April 2015
Cancer incidence rates have been rising throughout the world for several decades. No adequate explanation has been offered for this. A number of individual explanations have been offered, some of which concern factors which correlate quite well with individual cancer types. For example, ageing has been shown to correlate with hormonal cancers in Britain and with Bowel cancer worldwide, see for example but not exclusively, Peto (2001)  and Mohr and Feldman et al (2001) .
The present author has shown that at least in Britain, the sum of the incidence rates of the two main hormonal cancers, namely; breast and prostate cancer since 1947 correlates almost exactly with the number of TV Licenses issued and believes that both light and RF radiation are factors . Further in another study he has shown by using GIS data for GSM cell phone penetration, Night Light (wealth index) and Total Sunshine amounts that each of them are convoluted functions in correlation with incidence statistics for 10 common cancers on a global scale. He further emphasises the potency of Radio Frequency emissions as a larger potential driver of cancer by melatonin suppression then light, yet only the latter an acknowledged problem in many quarters .
The present author has also shown that at least in the USA, urban living is safer than country with regard to cancer except for Breast, Prostate and Kidney cancer incidence. Many states in the USA still permit the use of dangerous pesticides presently banned in the EU community.
Finally the present author has also shown that contrary to popular belief sunshine does more good than harm when it comes to cancer . Malignant melanoma only seems to be a problem in regions of the country and the world with a high radon concentration and low sunshine amounts in spring and autumn.
So what if there were to be a technological factor which could be more predictive of all cancers worldwide? Almost on a hunch, but with a degree of logical reasoning the present author decided to test the factor know as ‘Percentage Internet Penetration’ as a correlating factor for cancer incidence. The logic to this is as follows. Internet Penetration should be expected to be a measure of a country’s affluence, see for example Lucas and Sylla (2003). Following the argument aboves, more affluent countries ought to have more ageing and therefore more cancer. More affluent countries tend to produce more light at night and with internet comes more RF radiation, light from monitor screens and household lights after dark rather akin to television, therefore also more cancer. With internet penetration more physical inactivity is to be expected, more tendencies to stay indoors and suffer reduced vitamin D status hence through these multiple factors also we would expect more cancer. Compelling evidence exists for the relationship between physical inactivity and breast and bowel cancer . Inactivity is associated with TV viewing and internet use.
More affluent countries have populations which have a tendency to consume more junk/processed food, therefore also more cancer, see for example but not exclusively, Jakszyn et al (2006) . More affluent countries tend to fill their houses with furnishings made from modern materials leaking formaldehyde and the like, therefore more cancer, see Bosetti et al (2008) . More affluent nations, with the exception of affluent Islamic nations, will tend to consume more alcohol, therefore more cancer, particularly in conjunction with smoking. For example, at least 75% of head and neck cancers are attributable to a combination of cigarette smoking and alcohol drinking, see Hashibe, P Brennan, S Benhamou… - … the National Cancer …, 2007 - jnci.oxfordjournals.org .
All these separate, yet seemingly almost unquantifiable, effects will although having greater or lesser extents based on individual cancer types, be nevertheless convoluted together in the overall regression, with Cancer Incidence as Ordinate, versus Internet Penetration as abscissa. It is thus also to be expected therefore by way of this hypothesis that the more different cancer types (body sites) included the greater will be the regression factor.
Data on Internet Penetration was taken from:
http://www.internetlivestats.com/internet-users-by-country/ and data on world cancer incidence for twenty countries was taken from CRC UK. The nations used are ;Canada, USA, Mexico, Brazil, Argentina, UK, France, Germany, Poland, Russia, Saudi Arabia, Turkey, China, Nigeria, South Africa, Indonesia, Japan, Australia and India.
Results and Discussion
Linear regressions were plotted and calculated using CurveExpert by Hyams.
Figure 1 below shows a plot of all cancers incidence versus Internet Penetration.
A highly linear relationship is observed with a linear regression factor of .89. The two-tailed P value is less than 0.0001 By conventional criteria, this difference is considered to be extremely statistically significant. The implication is staggering. The suggestion is that almost 80% of cancers worldwide are associated with the factors discussed in the above hypothesis.
It is well worth mentioning that the largest outlier is the point at 52% Internet Penetration which belongs to Saudi Arabia. Here cancer incidence is significantly lower than predicted by the method. It is expected that potentially the strict laws on alcohol consumption may possibly account for some of the reduction in incidence and also at least in males a good vitamin D status is to be expected due to geographic position. It is hoped to report and a more extensive use of the method of using Internet Penetration correlation and specific residuals (outliers) to explore much more detailed cancer aetiology in the very near future.
It is instructive to consider the regression between Internet Penetration and Ageing (Life Expectancy), see Figure 2 below.
The regression value is .74. Therefore approximately 53% of the above 80% of cancers worldwide in the age range 53-86.2 are found in aging, i.e. 42%. This is in quite close agreement with the statistic specified by CRC UK for aging in the UK.
The initial hypothesis suggests that there should be no better correlation in this method than for all cancers combined. To test this and attempt has been made to correlate three other common cancers with Internet Penetration. These are breast, prostate and bowel cancers.
The result for breast cancer is shown in Figure 3 below.
Figure 3 Internet Penetration as a predictor of Breast Cancer R=.8
A similar correlation is shown for Prostate Cancer in figure 4 below.
Figure 4 Internet Penetration as a predictor of Breast Cancer R=.69
Figure 5 Internet Penetration as a predictor of Breast Cancer R=.79
As predicted no single cancer has a good a correlation with internet penetration as with all cancers combined although nevertheless highly significant correlation factors still exist with the above three named cancers. This strongly supports the notion of convoluted functionality in correlation and aetiology.
A final test of the hypothesis can be applied that is to see if the regression factor for several common cancers combined (but not all known cancers) increases over that for single cancers above. A regression of total world incidence of Lung, Prostate and Breast Cancer has been attempted against Internet Penetration, see Figure 6 below.
Figure 6 Internet Penetration as a predictor of totality of three common cancers R=.8
As expected, it is observed that the regression factor R=.8 is higher than any of the single cancers alone but less than for the totality of all cancers combined r=.89.
Once again Saudi Arabia ( 53% Internet Penetration) is the largest outlier with much lower rates of these three cancers combined than would be predicted by the model. Zero alcohol consumption and good vitamin D status particularly in males may account for this.
This evidence is strongly backed up by considering the comparative size of the Saudi outlier for the three cancers Breast, Prostate and Bowel.
The rank in order of lowest expected cancer probability ( largest negative outlier) with respect to the model is Prostate, Bowel, Breast.
This confirms the importance of vitamin D status in these three cancers. The notion is as follows. Saudi males get far more exposure to sunlight than Saudi females. As a result their Prostate cancer is far lower than expected. The bowel outlier is central because both males and females suffer from bowel cancer but only males will have the Vitamin D advantage. The breast outlier is so small as to be lost in the general scatter and indeed for breast cancer negative outliers from several other countries are comparable to or even outlying the one for Saudi Arabia, reflective of the general poor vitamin D status of the great majority of Saudi women who will not receive much sunlight directly to skin because of traditional/ religious dress code rules.
Percentage of Internet Penetration has been shown to be a highly accurate predictor of total cancer incidence worldwide. Further it has been possible to explore some aspects of cancer aetiology namely alcohol and vitamin D status. Moreover, it is hoped to report more extensively on this and other aetiologies by extending the present method in the very near future.
1. J Peto - Nature, 2001 - nature.com http://www.nature.com/nature/journal/v411/n6835/abs/411390a0.html
3. C. Barnes http://drchrisbarnes.co.uk/TVCAN.htm
4. C.Barnes http://drchrisbarnes.co.uk/CAGIS.htm
5. C. Barnes http://www.drchrisbarnes.co.uk/SUNGOOD.htm
7. Vainio H, Bianchini F, editors. IARC handbooks of cancer prevention Weight control and physical activity. Vol. 6. Lyon, France: IARC Press; 2000.